Diode Compact Models

1. Theory

• Definition: The SPICE diode model is the standard compact model of a pn or Schottky junction diode: an exponential current source with an emission coefficient, a series resistance, a depletion and diffusion charge, a reverse breakdown branch, temperature scaling and noise sources (MathWorks; Buffalo table; Modelica).
• Parameters: The core SPICE3 set has fifteen or sixteen parameters, IS, RS, N, TT, CJO, VJ, M, EG, XTI, KF, AF, FC, BV, IBV and TNOM, with the defaults most simulators share (Buffalo table; Modelica; All About Circuits).
• Extensions: PSpice-style models add a recombination current (ISR, NR), a high-injection knee (IKF), a second breakdown branch and temperature coefficients for RS, BV and IKF (MathWorks).
• Advanced models: For MOSFET source and drain junctions, JUNCAP2 splits the diode into bottom, STI-edge and gate-edge parts and adds SRH, trap-assisted and band-to-band tunnelling currents and avalanche breakdown (JUNCAP2 summary).

2. Working Principle

• Forward conduction: The intrinsic junction follows $I_S[\exp (V_D/(N{V}_T))-1]$; at high current the drop across RS and the high-injection factor bend the exponential (MathWorks; All About Circuits).
• Charge storage: The depletion charge follows the $(1-V/VJ{)}^{1-M}$ law and is continued linearly above FC times VJ to avoid the singularity at VJ, while the diffusion charge is TT times the current (MathWorks).
• Breakdown: A reverse exponential with slope set by NBV places the current IBV at the voltage BV, giving a sharp knee for zener and avalanche diodes (MathWorks; All About Circuits).
• Temperature: IS rises with temperature through XTI and EG, which makes the forward voltage at fixed current fall by about 2 $\mathrm{mV}$ per kelvin, and VJ and CJO shift with the band gap (MathWorks; computed).

3. Applications

• Discrete-diode models: Vendor and datasheet-derived cards for rectifiers, signal, Schottky, germanium and zener diodes (All About Circuits).
• Integrated junctions: Source and drain junctions of MOSFETs, well diodes and ESD diodes in design kits, using JUNCAP2 or SPICE diodes with area and perimeter scaling (JUNCAP2 summary; background knowledge).
• Circuit simulation: Rectifiers, clamps, mixers, detectors and reference circuits, where TT and CJO set switching and RF behaviour (All About Circuits; background knowledge).
• Temperature sensors and references: The forward-voltage temperature coefficient set by EG and XTI (MathWorks; computed).

DC Current

Name	Formula	Related Formulae	Variables/Constants
Shockley diode current with emission coefficient	$I_D=AREA\cdot IS\left[\exp \left(\frac{V_D}{N{V}_T}\right)-1\right],V_T=\frac{k_{B}T}{q}$	$V_{ext}=V_D+I_D{R}_S,R_S=\frac{RS}{AREA},IS\approx \frac{I_F}{\exp \left(V_F/(N{V}_T)\right)}$	$I_D:diodecurrent$ $AREA:areafactor$ $IS:saturationcurrent(default1\times 10^{-14}A)$ $V_D:voltageacrosstheintrinsicjunction$ $\begin{array}{r}N:emissioncoefficient(default1,\\ between1and2)\end{array}$ $V_T:thermalvoltage$ $k_B:Boltzmannconstant$ $T:temperature$ $q:elementarycharge$ $V_{ext}:terminalvoltage$ $RS:seriesresistance(default0)$ $\begin{array}{r}{I}_F,V_F:adatasheetforwardpoint;Buffalo\\ table,ModelicaandAllAboutCircuits:\\ thesamedefaults,withIS,RSandCJO\\ scaledbyarea;MathWorks:IS,IBVandCJO\\ multipliedandRSdividedbyAREAtimes\\ SCALE;AllAboutCircuits:Nofabout2\\ suggestedforpowerrectifiers\end{array}$ $\begin{array}{r}(computed:the1N4004pointof1\mathrm{A}at\\ 0.925\mathrm{V}withN=2and{V}_T=26\mathrm{mV}givesIS\\ =18.8\mathrm{nA},matchingAllAboutCircuits;at\\ 27{}^{\circ }C(V_T=25.86\mathrm{mV})thesamemodelgives\\ 1.10\mathrm{A}at0.925\mathrm{V})\end{array}$
Recombination and high-injection extension	$I_{fwd}=I_{nrm}{K}_{inj}+I_{rec}{K}_{gen},K_{inj}=\sqrt{\frac{IKF}{IKF+I_{nrm}}}$	$I_{rec}=ISR\left[\exp \left(\frac{V_D}{NR{V}_T}\right)-1\right],K_{gen}={\left[{\left(1-\frac{V_D}{VJ}\right)}^2+0.005\right]}^{M/2}$	$I_{fwd}:forwardcurrent$ $I_{nrm}:idealdiffusioncurrent$ $K_{inj}:high-injectionfactor$ $\begin{array}{r}IKF:high-injectionkneecurrent(default\\ infinite)\end{array}$ $I_{rec}:space-chargerecombinationcurrent$ $\begin{array}{r}ISR:recombinationsaturationcurrent\\ (default0)\end{array}$ $\begin{array}{r}NR:recombinationemissioncoefficient\\ (default2)\end{array}$ $K_{gen}:generationfactor$ $VJ:junctionpotential$ $\begin{array}{r}M:gradingcoefficient;MathWorks:these\\ PSpice-styletermswiththedefaults\\ listed,absentfromtheSPICE3parameter\\ listsoftheBuffalotableandModelica\\ (comparison,myreading)\end{array}$ $\begin{array}{r}(computed:K_{inj}=0.707at{I}_{nrm}=IKF;well\\ aboveIKFthecurrentgrowsasthesquare\\ rootof{I}_{nrm},sotheeffectiveemission\\ coefficientdoublesto2N)\end{array}$
Reverse breakdown	$I_{rev}=IBV\exp \left(-\frac{V_D+BV}{NBV{V}_T}\right)+IBVL\exp \left(-\frac{V_D+BV}{NBVL{V}_T}\right)$	$I_{rev}\left(V_D=-BV\right)=IBV+IBVL$	$I_{rev}:reversebreakdowncurrent$ $\begin{array}{r}BV:breakdownvoltage(positivenumber,\\ defaultinfinite)\end{array}$ $\begin{array}{r}IBV:currentatbreakdown(default1\times 10^{-3}A\\ inSPICE3lists),NBV\end{array}$ $NBVL:breakdownidealityfactors$ $\begin{array}{r}IBVL:low-levelbreakdowncurrent;\\ MathWorks:thesetwoexponentialswithIBV\\ default1\times 10^{-10}A/{\mathrm{m}}^2(seetheQ/A);Buffalo\\ table:BVinfinite,IBV1.0\times 10^{-3}(itsunit\\ columnsaysvolts);Modelica:BVdefault\\ -1\times 10^{40}standingforinfinityandIBV1\times 10^{-3}\\ A;AllAboutCircuits:BVandIBVfroma\\ datasheetleakagepoint(400\mathrm{V}and5\mu \mathrm{A}for\\ the1N4004),andazenermodelledby\\ settingBVtothezenervoltage\end{array}$ $\begin{array}{r}(computed:withNBV=1thereverse\\ currentrisesbyafactoreforevery{V}_{T}of\\ extrareversebias,sogoingfrom5\mu \mathrm{A}to1\\ \mathrm{mA}takesonly0.137\mathrm{V}beyondBVat27{}^{\circ }C)\end{array}$

Charge and Capacitance

Name	Formula	Related Formulae	Variables/Constants
Depletion capacitance with forward-bias linearization	$C_j=\frac{CJO}{{\left(1-V_D/VJ\right)}^M}(V_D<FC\cdot VJ),C_j=\frac{CJO}{F_2}\left(F_3+\frac{M{V}_D}{VJ}\right)(V_D\ge FC\cdot VJ)$	$F_1=\frac{VJ\left[1-{\left(1-FC\right)}^{1-M}\right]}{1-M},F_2={\left(1-FC\right)}^{1+M},F_3=1-FC\left(1+M\right)$	$C_j:depletioncapacitance$ $CJO:zero-biascapacitance(default0)$ $VJ:junctionpotential(default1\mathrm{V})$ $\begin{array}{r}M:gradingcoefficient(default0.5,\\ abrupt;about0.33forlinearlygraded)\end{array}$ $\begin{array}{r}FC:forward-biascapacitancecoefficient\\ (default0.5)\end{array}$ $\begin{array}{r}{F}_1,F_2,F_3:constantsofthelinear\\ continuation;MathWorks:thejunction\\ chargeforbothregionswith{F}_1,F_{2}and\\ F_3;AllAboutCircuits:M=0.333for\\ linearlygradedpowerrectifiers,0.5for\\ abruptjunctions,CJOreadfromthe\\ capacitancecurvenearzerobias(30\mathrm{pF}at\\ 1\mathrm{V}forthe1N4004);capacitanceformas\\ thederivativeofthecharge(background\\ knowledge)\end{array}$ $\begin{array}{r}(computed:forthe1N4148card(CJO=4\\ \mathrm{pF},M=0.285,VJ=0.6\mathrm{V}),C_j=2.12\mathrm{pF}at\\ -5\mathrm{V}and4.87\mathrm{pF}at{V}_D=FCVJ=0.3\mathrm{V};\\ bothbranchesgiveCJO/(1-FC{)}^{M}there,so\\ C_{j}iscontinuous)\end{array}$
Diffusion (transit-time) charge and capacitance	$Q_d=TT\cdot I_{fwd},C_d=TT\frac{d{I}_D}{d{V}_D}\approx \frac{TT\cdot I_D}{N{V}_T}$	$TT\approx \frac{Q_{RR}}{I_F}$	$Q_d:storedminority-carriercharge$ $TT:transittime(default0)$ $C_d:diffusioncapacitance$ $Q_{RR}:reverse-recoverycharge$ $\begin{array}{r}{I}_F:forwardcurrent;MathWorks:the\\ junctionchargeTTtimesAREAtimes{I}_{fwd}\\ plusthedepletioncharge;AllAbout\\ Circuits:TTestimatedas{Q}_{RR}/I_F,leftat0\\ whenthedatasheetgivesno{Q}_{RR},about4.32\\ \mu \mathrm{s}fora10A04rectifierand5\mathrm{ns}inthe\\ 1N4148card(C_{d}form,backgroundknowledge)\end{array}$ $\begin{array}{r}(computed:forthe1N4148card(TT=5\mathrm{ns},\\ N=1.24),C_d=1.56\mathrm{nF}at10\mathrm{mA}and27{}^{\circ }C\\ (V_D=0.624\mathrm{V}),about250timesthe6.4\mathrm{pF}\\ depletioncapacitancethere)\end{array}$

Temperature, Noise and Advanced Models

Name	Formula	Related Formulae	Variables/Constants
Temperature scaling of the saturation current	$IS\left(T\right)=IS\left(T_0\right){\left(\frac{T}{T_0}\right)}^{XTI/N}\exp \left[\left(\frac{T}{T_0}-1\right)\frac{qEG}{N{k}_{B}T}\right]$	$E_g\left(T\right)=1.16-\frac{7.02\times 10^{-4}{T}^2}{T+1108}\mathrm{eV},CJO\left(T\right)=CJO\left[1+M\left(4\times 10^{-4}\left(T-T_0\right)-\frac{VJ\left(T\right)-VJ}{VJ}\right)\right]$	$\begin{array}{r}{T}_0:parametermeasurementtemperature\\ (TNOM,default27{}^{\circ }C)\end{array}$ $\begin{array}{r}XTI:saturation-currenttemperature\\ exponent(default3forpnjunctions,2\\ forSchottky)\end{array}$ $\begin{array}{r}EG:activationenergy(default1.11\mathrm{eV}for\\ silicon,0.69forSchottkyand0.67for\\ germaniumintheAllAboutCircuits\\ cards),E_g(T):bandgapusedtoscaleVJ,\\ VJ(T)=VJT/T_0-3V_{T}ln(T/T_0)-E_g(T_0)T/T_0+\\ E_g(T);MathWorks:theseequations,plus\\ linearandquadraticcoefficientsforIKF,\\ BVandRS(TIKF,TBV1,TBV2,TRS1,TRS2);\\ BuffalotableandModelica:TNOM,EGand\\ XTI\end{array}$ $\begin{array}{r}(computed:N=1,XTI=3andEG=1.11\mathrm{eV}\\ giveariseof16.5percentperkelvin\\ andafactor4.40per10\mathrm{K}near300\mathrm{K};at\\ 1\mathrm{mA}withIS=1\times 10^{-14}Atheforwardvoltage\\ is0.655\mathrm{V}at27{}^{\circ }Candfallsby1.77\\ mV/K;E_g(300K)=1.115\mathrm{eV})\end{array}$
Diode noise sources	$\overline{i_n^2}=\left(2q{I}_D+\frac{KF\cdot I_D^{AF}}{f}\right)\Delta f$	$\overline{i_{R_S}^2}=\frac{4k_{B}T}{R_S}\Delta f$	$\begin{array}{r}\overline{i_n^2}:mean-squarenoisecurrentofthe\\ junction\end{array}$ $KF:flicker-noisecoefficient(default0)$ $AF:flicker-noiseexponent(default1)$ $f:frequency$ $\begin{array}{r}\Delta f:bandwidth,\overline{i_{R_S}^2}:thermalnoiseofthe\\ seriesresistance(forms,background\\ knowledge);Buffalotable,Modelicaand\\ AllAboutCircuits:KFandAFwiththese\\ defaults;MOSFETflicker-noiseformson\\ theMOSFETNoiseModelspage\end{array}$ $\begin{array}{r}(computed:at1\mathrm{mA}theshotnoiseis17.9\\ pA/\sqrt{\mathrm{Hz}};a10\Omega seriesresistanceadds40.7\\ pA/\sqrt{\mathrm{Hz}}at300\mathrm{K},butasacurrentintoa\\ short,soitseffectdependsonthesource\\ impedance)\end{array}$
JUNCAP2 geometric junction model	$I_j=MULT\left(AB{I}_{bot}^{\prime }+LS{I}_{sti}^{\prime }+LG{I}_{gat}^{\prime }\right),C_{jo}=CJOR{\left(\frac{VBIR}{V_{bi}}\right)}^P$	$I_{BBT}^{\prime }=CBBT{V}_{AK}{F}_{max}^2\exp \left(-\frac{FBBT}{F_{max}}\right),f_{bd}=\frac{1}{1-{\left(-V_{av}/VBR\right)}^{PBR}}$	$I_j:junctioncurrent$ $MULT:multiplier$ $AB:bottomarea$ $\begin{array}{r}LS,LG:STI-edgeandgate-edgeperimeters,\\ I^{\prime }:currentperareaorlengthofeach\\ component\end{array}$ $C_{jo}:zero-biascapacitanceattemperature$ $\begin{array}{r}CJOR:zero-biascapacitanceatthe\\ referencetemperature\end{array}$ $VBIR:built-involtageatreference$ $V_{bi}:built-involtageattemperature$ $\begin{array}{r}P:gradingcoefficient,I_{BBT}^{\prime }:band-to-band\\ tunnellingcurrent\end{array}$ $CBBT:prefactor$ $V_{AK}:anode-cathodevoltage$ $F_{max}:maximumjunctionfield$ $FBBT:normalizationfield$ $f_{bd}:avalanchemultiplicationfactor$ $V_{av}:effectivevoltage$ $VBR:breakdownvoltage$ $\begin{array}{r}PBR:onsetparameter;JUNCAP2summary:\\ ideal,SRH,trap-assistedtunnellingand\\ band-to-bandtunnellingcurrentstimesthe\\ breakdownfactor,single-pieceSRHandTAT\\ expressionsvalidinforwardandreverse\\ bias,arbitrarygradingcoefficient,\\ defaultsCJORBOT1\times 10^{-3}F/{\mathrm{m}}^2,VBIRBOT1\mathrm{V},\\ PBOT0.5(maximum0.95),IDSATRBOT1\times 10^{-12}\\ A/{\mathrm{m}}^2,VBRBOT10\mathrm{V},PBRBOT4\mathrm{V},FREV1\times 10^3\\ andareferencetemperatureof21{}^{\circ }C\end{array}$ $$\begin{gathered} \begin{array}{r}(computed:CJORBOT=1\times 10^{-3}F/{\mathrm{m}}^{2}is1fF/\mu \\ {\mathrm{m}}^2)\end{array} \end{gathered}$$

Important Questions

Ques: How are the SPICE diode parameters extracted from a datasheet?

All About Circuits works through a 1N4004 rectifier. N is chosen as 2 for a power rectifier; IS then follows from one point of the forward curve, 1 $\mathrm{A}$ at 0.925 $\mathrm{V}$, which with $V_T=26$ $\mathrm{mV}$ gives 18.8 $\mathrm{nA}$ (reproduced, computed). CJO is read from the capacitance curve near zero bias, 30 $\mathrm{pF}$ at 1 $\mathrm{V}$; M is set to 0.333 for a linearly graded junction; BV and IBV come from the leakage specification of 5 $\mu \mathrm{A}$ at 400 $\mathrm{V}$. TT would be $Q_{RR}/I_F$, but no recovery charge is given, so it stays at zero. Finally RS is tuned against the high-current part of the curve, where 28.6 $\mathrm{m\Omega }$ brings the 1.4 $\mathrm{V}$ current to about 12 $\mathrm{A}$. Real extractions fit IS and N to the whole low-current curve. Reference: ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary).

Ques: Does the datasheet 1N4004 model reproduce the currents the article reports?

Only partly. With IS = 18.8 $\mathrm{nA}$, N = 2 and RS = 0, a pure exponential gives 1.10 $\mathrm{A}$ at 0.925 $\mathrm{V}$ at the 27 ${}^{\circ }\mathrm{C}$ SPICE default (computed), matching the article. At 1.4 $\mathrm{V}$, however, the same model gives about 10.7 $\mathrm{kA}$ (computed), not the 16.2 $\mathrm{A}$ the article reports; 16.2 $\mathrm{A}$ needs about 21 $\mathrm{m\Omega }$ of resistance in the path (computed), so the simulated circuit probably had some series resistance or the value was misreported (my reading). With RS = 28.6 $\mathrm{m\Omega }$ the computed currents at 27 ${}^{\circ }\mathrm{C}$ are 12.2 $\mathrm{A}$ at 1.4 $\mathrm{V}$ and 0.73 $\mathrm{A}$ at 0.925 $\mathrm{V}$, reproducing the article’s values exactly. The article’s card also writes CJO=30 without a unit suffix, which SPICE reads as 30 farads, not 30 $\mathrm{pF}$ (background knowledge for SPICE suffixes). Reference: ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary).

Ques: Why does the SPICE depletion capacitance switch to a linear form above FC times VJ?

Because $CJO/(1-V_D/VJ{)}^M$ diverges at $V_D=VJ$, which a forward-biased diode easily reaches. MathWorks gives the SPICE remedy: below FC times VJ the charge follows the power law, and above it a quadratic charge, with constants $F_1$, $F_2=(1-FC{)}^{1+M}$ and $F_3=1-FC(1+M)$, makes the capacitance a straight line. Both forms give $CJO/(1-FC{)}^M$ at the boundary (computed), so charge and capacitance stay continuous. For the 1N4148 card, with CJO = 4 $\mathrm{pF}$, M = 0.285 and VJ = 0.6 $\mathrm{V}$, the capacitance is 2.12 $\mathrm{pF}$ at -5 $\mathrm{V}$, 4.87 $\mathrm{pF}$ at the 0.3 $\mathrm{V}$ boundary and 6.4 $\mathrm{pF}$ at 10 $\mathrm{mA}$ (computed). There the diffusion capacitance of 1.56 $\mathrm{nF}$ dominates by about 250 times (computed), so the linear form only needs to be smooth. Reference: MathWorks, ‘SPICE Diode’, Simscape Electrical block reference page (fetch-tool summary); ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary).

Ques: Is the default breakdown current IBV 1 mA or $1\times 10^{-10}$ A per square metre?

The two describe different conventions. The Buffalo table, Modelica’s SPICE3 modelcard and All About Circuits all give IBV a default of $1\times 10^{-3}$ $\mathrm{A}$ (the Buffalo table’s unit column says volts, a slip for amperes), the SPICE2 and SPICE3 value. MathWorks lists $1\times 10^{-10}$ $A/{\mathrm{m}}^2$ because its block scales IBV with an area in square metres and uses a softer two-exponential breakdown. For a diode of AREA = 1, as in a plain SPICE card, the MathWorks default is seven orders of magnitude lower (computed), so a card moved between simulators without IBV puts the breakdown knee at a different current. With NBV = 1 the knee moves by $V_T\ln (10^7)=0.42$ $\mathrm{V}$ at 27 ${}^{\circ }\mathrm{C}$ (computed). The SPICE3 value is more likely the intended default for netlists, and BV and IBV should always be set explicitly (my reading). Reference: ‘SPICE diode table’, SPICE pn-diode applet pages, University at Buffalo (author not stated) (fetch-tool summary); Modelica Association, Modelica Standard Library 4.0.0, ‘Modelica.Electrical.Spice3.Semiconductors.ModelcardDIODE’ (fetch-tool summary); ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary); MathWorks, ‘SPICE Diode’, Simscape Electrical block reference page (fetch-tool summary).

Ques: How strongly does the saturation current depend on temperature?

Very strongly. MathWorks gives $IS(T)=IS(T_0)(T/T_0{)}^{XTI/N}\exp [(T/T_0-1)qEG/(N{k}_{B}T)]$. With N = 1, XTI = 3 and EG = 1.11 $\mathrm{eV}$, IS rises by 16.5 per cent per kelvin near 300 $\mathrm{K}$ and by a factor 4.40 over 10 $\mathrm{K}$ (computed), close to the familiar doubling every 5 $\mathrm{K}$. At a fixed 1 $\mathrm{mA}$ the forward voltage, 0.655 $\mathrm{V}$ for IS = $1\times 10^{-14}$ $\mathrm{A}$ at 27 ${}^{\circ }\mathrm{C}$, therefore falls by 1.77 $mV/K$ (computed). Schottky diodes use XTI = 2 and EG of about 0.69 $\mathrm{eV}$, as in the All About Circuits cards for the 1N5711 and 1N5712, and germanium 0.67 $\mathrm{eV}$. VJ and CJO scale too, through a band gap that MathWorks takes as 1.16 minus $7.02\times 10^{-4}{T}^2/(T+1108)$ $\mathrm{eV}$, 1.115 $\mathrm{eV}$ at 300 $\mathrm{K}$ (computed). Reference: MathWorks, ‘SPICE Diode’, Simscape Electrical block reference page (fetch-tool summary); ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary).

Ques: Does the MathWorks page apply the emission coefficient twice?

As read, yes. The extract defines $V_t=N{k}_{B}T/q$ and then writes the diffusion current as $IS[\exp (V_d/(N{V}_t))-1]$, which would put $N^2$ in the exponent, and the IS temperature law also divides by $N{V}_t$. Every SPICE3 parameter list, including the Buffalo table and Modelica, treats N as a single emission coefficient in $\exp (V_D/(N{k}_{B}T/q))$ (background knowledge for the SPICE3 form), and the All About Circuits extraction of the 1N4004 uses $V_T=26$ $\mathrm{mV}$ without N inside. For the 1N4004 card with N = 2, double counting would multiply the exponent’s denominator by two again and give 0.14 $\mathrm{mA}$ instead of 1.10 $\mathrm{A}$ at 0.925 $\mathrm{V}$ (computed). The single-N form is almost certainly intended, and the extract’s $V_t$ is probably the plain thermal voltage (my reading). Reference: MathWorks, ‘SPICE Diode’, Simscape Electrical block reference page (fetch-tool summary); ‘SPICE diode table’, SPICE pn-diode applet pages, University at Buffalo (author not stated) (fetch-tool summary); Modelica Association, Modelica Standard Library 4.0.0, ‘Modelica.Electrical.Spice3.Semiconductors.ModelcardDIODE’ (fetch-tool summary); ‘SPICE Models’, Lessons in Electric Circuits, Volume III (Semiconductors), Chapter 3, All About Circuits (author not named on the page) (fetch-tool summary).

Ques: What does JUNCAP2 add over the SPICE diode for MOSFET junctions?

Geometry and leakage physics. JUNCAP2 splits each junction into bottom, STI-edge and gate-edge components, scaled by area AB and perimeters LS and LG, each with its own parameters, so one card fits all layouts. Besides the ideal diffusion current it includes Shockley-Read-Hall generation, trap-assisted tunnelling and band-to-band tunnelling, $CBBT{V}_{AK}{F}_{max}^2\exp (-FBBT/F_{max})$, multiplied by an avalanche factor $1/[1-(-V_{av}/VBR{)}^{PBR}]$; the SRH and TAT terms are single-piece expressions valid in forward and reverse bias. The capacitance allows any grading coefficient up to 0.95, with a default of $1\times 10^{-3}$ $F/{\mathrm{m}}^2$, or 1 $fF/\mu {\mathrm{m}}^2$ (computed). These terms matter for off-state leakage of scaled MOSFETs with heavily doped halos, where band-to-band tunnelling dominates (background knowledge). JUNCAP2 is the junction model used with PSP. Reference: A. J. Scholten, M. Durand, G. D. J. Smit, D. B. M. Klaassen, O. Rozeau, S. Martinie and T. Poiroux, ‘JUNCAP2 200.6.2’ model summary, NXP Semiconductors and CEA-Leti (April 2021) (fetch-tool summary).

References

1. MathWorks, “SPICE Diode,” Simscape Electrical documentation, block reference page. Available: https://www.mathworks.com/help/sps/ref/spicediode.html
2. “SPICE diode table,” SPICE pn-diode applet pages, University at Buffalo (author not stated). Available: https://www.acsu.buffalo.edu/~wie/applet/spice_pndiode/spice_diode_table.html
3. Modelica Association, “Modelica.Electrical.Spice3.Semiconductors.ModelcardDIODE,” Modelica Standard Library 4.0.0 documentation. Available: https://doc.modelica.org/Modelica%204.0.0/Resources/helpWSM/Modelica/Modelica.Electrical.Spice3.Semiconductors.ModelcardDIODE.html
4. “SPICE Models,” Lessons in Electric Circuits, Vol. III: Semiconductors, Ch. 3, All About Circuits (author not named on the page). Available: https://www.allaboutcircuits.com/textbook/semiconductors/chpt-3/spice-models
5. A. J. Scholten, M. Durand, G. D. J. Smit, D. B. M. Klaassen, O. Rozeau, S. Martinie and T. Poiroux, “JUNCAP2 200.6.2,” model summary, NXP Semiconductors and CEA-Leti, Apr. 2021. Available: https://www.cea.fr/cea-tech/leti/pspsupport/Documents/Level%20104/juncap200p6_summary.pdf

Related pages in this vault: Ebers-Moll and Gummel-Poon Models, Temperature Scaling, Parameter Extraction Methods, Power Device Compact Models, PN Junction, Zener Diode, Backward Diode, Avalanche Diode, Schottky Diode, PIN Diode, GaN Power Diode, SiC Schottky Diode, Step Recovery Diode, Point Contact Diode.